|
--- |
|
license: apache-2.0 |
|
tags: |
|
- summarization |
|
- persian |
|
- generated_from_trainer |
|
datasets: |
|
- xlsum |
|
model-index: |
|
- name: mt5-base-finetuned-persian |
|
results: [] |
|
--- |
|
|
|
<!-- This model card has been generated automatically according to the information the Trainer had access to. You |
|
should probably proofread and complete it, then remove this comment. --> |
|
|
|
# mt5-base-finetuned-persian |
|
|
|
This model is a fine-tuned version of [google/mt5-base](https://huggingface.co/google/mt5-base) on the xlsum dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 3.6086 |
|
- Rouge-1: 22.02 |
|
- Rouge-2: 7.41 |
|
- Rouge-l: 18.95 |
|
- Gen Len: 19.0 |
|
- Bertscore: 69.89 |
|
|
|
## Model description |
|
|
|
More information needed |
|
|
|
## Intended uses & limitations |
|
|
|
More information needed |
|
|
|
## Training and evaluation data |
|
|
|
More information needed |
|
|
|
## Training procedure |
|
|
|
### Training hyperparameters |
|
|
|
The following hyperparameters were used during training: |
|
- learning_rate: 0.0005 |
|
- train_batch_size: 4 |
|
- eval_batch_size: 4 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 8 |
|
- total_train_batch_size: 32 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- num_epochs: 5 |
|
- label_smoothing_factor: 0.1 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Rouge-1 | Rouge-2 | Rouge-l | Gen Len | Bertscore | |
|
|:-------------:|:-----:|:----:|:---------------:|:-------:|:-------:|:-------:|:-------:|:---------:| |
|
| 7.2823 | 0.96 | 19 | 3.9800 | 19.78 | 5.57 | 16.24 | 19.0 | 68.19 | |
|
| 4.7334 | 1.96 | 38 | 3.7620 | 20.92 | 7.49 | 18.27 | 18.91 | 68.72 | |
|
| 4.3891 | 2.96 | 57 | 3.6349 | 21.07 | 7.66 | 18.53 | 18.96 | 69.73 | |
|
| 4.2 | 3.96 | 76 | 3.6315 | 19.63 | 6.49 | 16.61 | 19.0 | 69.15 | |
|
| 3.9202 | 4.96 | 95 | 3.6086 | 21.2 | 6.8 | 17.06 | 19.0 | 69.48 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.18.0 |
|
- Pytorch 1.11.0+cu113 |
|
- Datasets 2.1.0 |
|
- Tokenizers 0.12.1 |
|
|
